ABSTRACT Every possible missense variant that is compatible with life is likely present in the germline of a living human. Some of these variants alter protein activity or abundance, and, consequently, may impact disease risk. However, only ~2% of all presently reported missense variants have clinical interpretations. Most of the remaining variants, as well as nearly all missense variants not yet observed, are rare and cannot be interpreted using traditional approaches, creating a major challenge for the clinical use of genomic information. Our goal is to address this challenge by measuring the functional consequences of nearly every possible missense variant in clinically relevant proteins using deep mutational scanning. In a deep mutational scan, a library of protein variants is subjected to selection for the function of the protein, and high-throughput DNA sequencing is used to read out the enrichment or depletion of each variant, revealing the variant's function. Despite recent progress, deep mutational scanning suffers from two major limitations. The first lies in the requirement to handcraft a specific assay for the function of each protein. With over 4,000 disease-associated genes in the human genome, this one-at-a-time approach is impractical. Thus, we propose Variant Abundance by Massively Parallel Sequencing (VAMP-seq), a functional assay that is both informative of variant effect and generalizable to many proteins. The assay is based on the fact that, despite their diversity, most proteins share a key requirement: they must be abundant enough to perform their molecular function. We will generate VAMP-seq abundance data for nearly all possible missense variants in a set of ten clinically important proteins, refining VAMP-seq as a tool for assessing missense variation in many, if not most, disease-relevant genes. We will also combine VAMP-seq with chemical perturbations to reveal fundamental features of protein synthesis, folding and degradation, as well as to identify variants whose low abundance could be ameliorated pharmacologically. The second major limitation is that deep mutational scans typically quantify the effect of variants on a protein's activity or on cell growth. These simple measurements sometimes fail to capture the complexity of the relationship between genotype and human phenotype. Thus, we propose Microscope- Assisted Visuospatial Sorting (MAViS), which will enable multiplex assessment of variant effects on more complex phenotypes like a cell's internal organization, shape or behavior. We will apply MAViS to several disease-related genes, generating rich phenotypic data for nearly all possible missense variants. The data we gather from both VAMP-seq and MAViS will be used to generate comprehensive ?look-up tables? describing the effects of nearly every missense variant in each gene. We will also analyze these variant effects in the context of known pathogenic and benign variants, using a learning-based approach to make comprehensive predictions of missense variant pathogenicity.